The best argument I have ever seen against the idea that humans are dangerously warming the earth — that is, against the view of Al Gore, Elizabeth Kolbert, and thousands of other people who claim to understand what they are talking about — comes, strangely enough, from a supporter of this view.
Steve Connor is the Science Editor of The Independent, a highbrow London newspaper. He interviewed Freeman Dyson — who, like me, thinks the conventional certainty on this issue is far too strong — on the subject. The headline of the interview labels Dyson a “heretic”. Connor wants to know how Dyson reacts to what seems to Connor to be overwhelming evidence.
The interview is by email. Dyson says he has no faith in the models. Connor writes:
I was only trying to find out where your problem lies with respect to the scientific consensus on global warming. As you know these models [that Dyson doesn’t believe] are used by large, prestigious science organizations such as NASA, NOAA and the Met Office, which use them to make pretty accurate predictions about the weather every day. The scientists who handle these models point out that they can accurately match up the computer predictions to real climatic trends in the past, and that it is only when they add CO2 influences to the models that they can explain recent global warming.
There it is. The scientists who use the weather models every day, who know them better than anyone else say that we should believe them because 1. They can fit “real climatic trends in the past”. This is meaningless. The models have lots of adjustable parameters. Perhaps they could have fit any plausible past trends. 2. They “make pretty accurate predictions about the weather every day” — that is, predictions of the weather of the next week or so.
This is admission of defeat. It’s as if you say you can throw a ball a mile and, when someone asks how you know this, you say, “I’ve thrown a ball 10 yards quite often.” If you had thrown a ball more than 10 yards you would have said so. If the models had predicted accurately more than a week in advance their boosters would have said so.
It isn’t just Steve Connor who unintentionally makes a really good case for the opposite of what he believes. Sir Paul Nurse, a Nobel Prize winner in Biology and President of the Royal Society, hosted a recent BBC show called Science Under Attack in which we were supposed to believe predictions of global catastrophe because weather models can predict the weather for the next few days. A NASA weather expert said that! Nurse took him seriously.
My goodness. If the President of the Royal Society is this credulous, what are the ordinary members like?
There is always a huge disconnect between science and the “scientists” who have access to the mainstream media.
Bravo Seth, a great use of your BS-detector and also very insightful about modeling in general. And ballsy to skewer ad sacred cow.
Questions from an American pragmatist:
Please unpack your sentence “who, like me, thinks the conventional certainty on this issue is far too strong”:
1. What level of confidence do you hold that those who argue that humans are affecting climate change are correct in some significant measure?
2. Are you saying those who contend that human action currently influences our weather and climate have the balance of evidence in their favor but you want to take no response? You want to wait and see? What is the potential risk of waiting? (Of course, there are real costs in acting, too.)
3. Given that Al Gore, among others, is only a layman who serves to raise this issue, do legitimate scientists as a whole back the Dyson-Roberts skeptical position? How should a layman, like me, respond?
4. Newton had some very odd beliefs (by modern standards) that seem at odds with his revolutionary physics, so is this a similar case with the otherwise incredibly intelligent Dyson? Can one who produces diamonds also produce dung?
5. Pragmatism uses skepticism, but it does not enshrine it upon an alter, lest it become cynicism or mere contrariness.
Hi Seth,
This is an honest question, not a snarky remark.
The main reason that I believe in human-driven global warming is that, after two hundred years of increased burning of hydrocarbons, putting lots and lots of CO2 into the air, wouldn’t you expect global warming? Or, how would you avoid this?
I assume that the cause-and-effect looks different from your end, and I’m curious how.
Thanks.
If this is the best argument against man made global warming then it declaring it a certainty.
That someone that does not work in climatology or weather, and does not know the difference between climate models and weather models doesn’t believe their accuracy is the best counter argument is great news for actual scientists working in the field.
And stating that weather models cannot predict climate does not change any of the measurable facts of CO2 increases or warming, or that CO2 has green house affects.
I think you raise a good point in that you can’t say a model is valid because it predicts the weather. But what would you have to see in a model to convince you that it is valid?
Your throwing the ball analogy doesn’t really hold up because we don’t have the option of throwing the ball a mile. We just have to wait 100 or 1000 years and see if we were right. I think the best scientists can do is create a model incorporating as much as we know about climate, and hope that it is correct. If that is true, then unfortunately the only people who are in a position to evaluate the model are those who have studied it in great detail. Which brings you right back to trusting the scientists because they are scientists. It’s a conundrum.
Tim
I read the other planets in our solar system change drastically for decades. I suggest David Wilcock and his research.
No cars on Mars as far as we know.
I’m in favor of not using cars and treating the earth well, btw. So that is not at issue. It’s about understanding the real cause, which seems to involve at least our whole solar system.
Well, if there’s one thing humans are notoriously bad it, it’s making predictions about future events. The more complex and chaotic the system, like the stock market, the harder to make predictions. Climate is a very complex, chaotic system, involving a wide variety of fields spanning from geophysics, meteorology, astronomy, and “climatology”. And the climatologists don’t really have any insight into the other fields. They are looking at a narrow set of variables rather than the true cross-discipline approach that we need to fully understand longterm climate variability.
Roy Spencer is a meteorologist who makes a strong claim that Pacific Decadal Oscillations–Pacific ocean temperature fluctuations which operate on 30 year cycles–can explain about 75% of the climate shift in the 20th century. He speculates that the PDO causes a change in cloud cover which causes radiative imbalance, causing longterm temperature shifts. This is a much more plausible explanation than blaming it all on CO2.
The IPCC models don’t even take into account the possibility that clouds can be primary drivers of climate change, since studying clouds is more of a meteorology discipline. So there’s a lot we don’t know about climate, and until we can take a truly cross-discipline approach, we will remain in the dark.
Clint, someone who works at NASA told Paul Nurse we can trust climate/weather models because they can predict the weather a few days later. I am not making this up. The person at NASA works in the field.
Tim Newsome, what I want to see to convince me a model is valid is accurate prediction. The models used by Jim Hansen et al. to predict catastrophe could easily have been used to predict the weather/climate 1 year in advance. Or 2 years in advance. They are what, 20 years old? If that had happened, now we would have a decent set of predictions to compare against what actually happened. But, as you may have noticed, this hasn’t been done or at least the results haven’t been made public. So people like Steve Connor are reduced to talking about predicting Thursday’s weather on Monday.
Calvin, adding more carbon dioxide to the air has lots of effects. I don’t know which ones matter the most nor how they are related. As the planet gets warmer for other reasons, carbon dioxide is released from the ocean. There is nothing new about big increases in atmospheric carbon dioxide. It hasn’t led to catastrophe in the past. Models that say large increases in atmospheric carbon dioxide must cause catastrophic warming are wrong.
Sean, good points. I agree.
Steve G., I believe that the Cassandras, such as Hansen, are placing too much weight on models that have never been shown to predict accurately. There is no good reason to believe those models. In psychology, the same thing happened: People took seriously models that had never been shown to accurately predict. Perhaps they confused fitting with prediction, I don’t know. Maybe Hansen and other climatologists confuse the two — I don’t know. But I do know that it is far harder to test a model’s predictions than to test whether it can fit past data. Have humans significantly affected the climate? I don’t know. I support research to find out. Which means testing climate models, among other things.
One scientist who has wavered in his views about anthropogenic global warming is Matt Ridley. I highly recommend his book ‘The Rational Optimist’ where he explains his current (skeptical) views.
I consider this a situation where the ‘alarmist’ crowd has sabotaged their own goals. It just feels so good and rightous to be against Big Oil and the Rich and to be one of those in-the-know as to the next looming catastrophe. And yet they haven’t thought this through. How could their imagined dystopia be prevented? By replacing fossil fuels with renewables. How does one generate political enthusiasm for renewables? The liberal crowd is already there, it’s the conservative crowd which needs to throw their weight behind the effort. Which could be done by appeals to patriotism, nationalism, self-reliance, and refusal to fund radical Islam (indirectly) through petro-dollars. Structure it not as an unlimited federal spending spree on research, but structure it to appeal to the capitalists, something along the lines of: The United States Government will buy $20 billion dollars of renewable energy technology each year for the next ten years (spread amongst the top ten most efficient American-created alternative-energy products). A guaranteed market which opens the funding for the building of factories. State governments could be encouraged to make similar pledges. The airwaves could be filled with slogans about ‘Self-Reliance in Ten Years’ and various others jingoistic ones having to do with camel racing.
Not that the conservatives have been bright enough to figure this one out either. It’s just been too satisfying on their side to whine about welfare queens or birth certificates or huge debts (which they conveniently forget were run up by their own leaders).
Read Matt Ridley. His optimism sometimes almost overwhelms my disgust with American politics.
Matt Ridley is a smart guy, and the Rational Optimist is a great book, and his arguments are probably on point, but the fact is that alarmism sells and optimism doesn’t. Alarmists can claim to be “informed” and “conscious” whereas people like Ridley are written off as naive, or “free market ideologues”. They don’t even need to argue against his points to write him off, they can just draw an unrelated comparison like how unrestrained capitalism created the recent financial crisis, and just like that, 99% of people would buy that argument. People like Ridley don’t have any influence in this debate, because he’s not telling people what they want to hear, whereas people like Al Gore are.
For Bruce M Cattanach BSc PhD DSc FRS rose-colored glasses triumph over alleged alarmism:
https://www.astraean.com/borderwars/2011/02/inbred-mistakes-vi.html#comments
Domestic animal breeding apparently being an animal quite different from climate change.
I believe that you are not making up any of this.
However, I think the point of such a statement is not “we can use the weather model to predict future climate” but more along the lines of “we used to be really bad at modeling weather and we have learned new and better ways of modeling and therefore we are fairly confident in our ability to create climate models”.
If anyone is seriously making the previous claim then they do not know what they are speaking about. And climate history has a bad history of people who are massively qualified in fields other than climatology making ignorant claims about climate science.
You say “They can fit “real climatic trends in the past.” This is meaningless. The models have lots of adjustable parameters. Perhaps they could have fit any plausible past trends” but what you’re criticizing includes the specific claim that “that it is only when they add CO2 influences to the models that they can explain recent global warming”. Do you mean to say that you consider this to be a lie? If so, it’s an odd one, where they (Gavin Schmidt in particular) seem to be working hard to make it easy for you to expose the problem: you can download the Goddard model from https://www.giss.nasa.gov/tools/modelE/ and the documentation is careful to explain how you can play with the parameters yourself. If you (or a group you can put together, perhaps?) can get the model to match past trends without CO2 influences, or even with a substantially reduced CO2 sensitivity, then you’ll really have something interesting. And surprising, at least to me.
I don’t know climate models from calamari, but I do think that scientific consensus is over-rated. There is fairly widespread consensus that mental illness responds well to treatment with psychotropic drugs (antidepressants, antipsychotics, etc.). However, I think the truth is that these drugs are dangerous and ineffective.
Seth, and everyone,
I think that my main point isn’t necessarily CO2. I’m not a scientist, and if I was, I don’t think I would be a climate scientist, just because that’s not my interest. In fact, I think that most of the comments in this thread are just a lot of lay people talking about what they’ve read in books and online. Me included.
That being said, I don’t think that the impact humanity has made on the world, the atmosphere included, is a small thing. There are a lot of things added, and a lot of things taken away. A lot more plastic in the ocean, a lot less uncovered soil on the ground, etc.
I think that it’s so easy to get behind an idea like global warming (whether it’s true or not, I honestly can’t say I’m certain) because we know, intuitively, that when something is driven more and more off-balance, it generally will hit a tipping point, and then change relatively quickly to a new balance. Avalanches, Jenga-towers, storms, the straw that broke the camel’s back, etc.
So my core belief is that we have changed the world a lot, and that changes in nature are prone to the domino effect. I also believe that, at some point, one of these changes, or a consequential change, won’t be tolerable to humanity, and that will be that. Whether or not global warming is involved, I guess, will not really be important when we get to that point.
Tom, no I don’t consider that “the only way they can fit recent warming…” to be a lie. Perhaps the models are missing X, and X causes both (a) carbon dioxide increases and (b) warming.
It’s extremely easy to make that mistake.
Models can be useful, and do have to be calibrated. But the calibrations are necessarily ambiguous – as you tune the model to play well with past measurements, various judgment calls have to be made, and if they’re made by someone who already believes in carbonic warming, those judgments are going to implicitly assume carbonic warming is occurring.
Essentially the models are circular arguments (well, probably) even setting aside all their other shortcomings.
My issue is that it appears that the model authors are aware of the problem, and for it is for this reason among others that they refuse to share the model.
You can’t and shouldn’t jail someone for pleading the fifth, but it should make you very very suspicious.
Seth, certainly it’s a possibility that all of the GCM models are missing something important; I don’t understand how what they’re missing could be an X that causes carbon dioxide increases. A GCM model isn’t really for predicting carbon dioxide increases anyway; the basic carbon dioxide level in the models is an input. For the postdictions (matching to past data) you were criticizing, that input is what was measured as being in the atmosphere (and sometimes what was measured as being in the ocean, as well.) We know, pretty closely, where the CO2 has been coming from ( https://en.wikipedia.org/wiki/List_of_countries_by_carbon_dioxide_emissions ), with a little less than 1% addition by volcanoes, on average.
There’s plenty of uncertainty left in the modeling, especially (as I understand it) about cloud formation, heat transfers within the ocean, and such. Because of that, you need to use parameters to put in assumptions, as well as parameter that simply represent inputs. But the modelers have been claiming, as you quoted, that they can make it work (postdiction, matching not only global averages but rough spatial distribution) if they assume the overall CO2 sensitivity that they do assume, somewhere in the range of 2-4.5C per doubling of CO2 (see, e.g. https://www.skepticalscience.com/Hansen-1988-prediction-advanced.htm ), and they can’t make it work without that. A lot of people have criticized their work, but apparently without being able to construct successful postdictions that didn’t assume roughly the same CO2 sensitivity. So we have a phenomenon for which we have successful models, but only on that assumption (or range of assumptions, if you prefer.) To me, that’s a pretty good test; it doesn’t make the President of the Royal Society look credulous.
Personally I made fun of global warming models in the early 80s, when as an asst prof of computer science I learned a bit about them; they deserved it then. (I followed Roy Spencer’s satellite-data adventures for some years thereafter, since which I’ve been inclined to ignore him or at least wait for refereed publications and criticisms thereof.) The models are certainly not perfect, but it has been a while since I made fun of them.
(But I’m still not worried about global warming in the long run; five years ago I wrote https://mistakesbytjm.blogspot.com/2005/11/long-run-global-warming-and-other.html and I’m inclined to stick with a surprisingly large fraction of it.)
alrenous, I’m surprised that you think “the model authors are aware of the problem, and for it is for this reason among others that they refuse to share the model.” As I said above, you can download the Goddard (NASA) model from https://www.giss.nasa.gov/tools/modelE/ and play with it yourself. I’m not sure how many years that’s been true, but it’s not new: I remember recommending it to a computer science student a couple of years ago (I’m not a prof any more, just a grant-funded programmer who sometimes works on projects that comp-sci students are also working on.) There used to be complaints that it was hard to get the model going, but there now seems to be a community of people doing it. There are other models you can play with as well; see the last question on https://www.realclimate.org/index.php/archives/2008/11/faq-on-climate-models/ . Enjoy!
1. Kirk: Matt Ridley isn’t a scientist. He’s a journalist and hereditary plutocrat. His doctorate in zoology, and marriage to a neuroscientist, do not bolster his opinions on climate. Prophetically, in 2007 he was forced out of chairmanship of the Northern Rock bank — the first of UK’s banks to collapse since 1878! — for failing to heed warnings about its unwise policies. Come 2008, he had lots of company in ignominy, which in no way reduces his own culpability or theirs. His sociobiological fancies haven’t fared better than his bank. He does not appear to have taken any lesson from his failures.
2. It’s curious that Seth refers to climate scientists as “Cassandras”. Cassandra is best remembered for having been right, but ignored. It was, precisely, ignoring the original Cassandra that brought down disaster on her kin.
3. By the same logic Seth presents, his posting demonstrates the weakness of the denialist argument. The strongest, most persuasive bit of evidence you can muster is an offhand remark, in an interview question, that, examined, turns out to be completely neutral on the topic?
4. For Alrenous, re: “they refuse to share the model”: as Tom (no relation) points out, the model is readily available; go wild. This, by the way, completely refutes your assertion.
5. For Alex: I was off work for 10 weeks last year with a short-term-memory disability. (Thanks be to our recent ancestors for establishing state Disability Insurance; without, my family would be in penury.) Once the right dangerous and ineffective psychotropic drug was identified, I was back at work within days. To convince me they never work, you’ll need much stronger evidence than you can muster.
6. Finally: https://scienceblogs.com/startswithabang/2011/02/weekend_diversion_jewels_of_th.php
“As I said above, you can download the Goddard (NASA) ”
Yeah, embarassed myself pretty good there.
I even saw your comment – after it was too late. Learn not to post when sleepy, I guess…?
Nathan, I wouldn’t say that psychotropic drugs never work. Even a bullet to the head works sometimes, as evidenced by this case report of a man who successfully treated his OCD by shooting himself.
Also, I wouldn’t consider memory deficits to be a mental illness per se. Sounds more like a neurological problem. In any case, the only drugs that are typically prescribed for treating memory deficits (at least in the US) are the drugs used for dementia — and even many doctors view those drugs as useless (they’re essentially placebos-by-proxy — they are designed to make the caregivers feel better).
Nathan, okay Matt Ridley’s Ph.D. is in zoology. Here’s what I’d like to know: You seem to take climate models seriously. Why?
Alex: I was allowed to visit a neurologist, too, who explained (among other things) that my problem was certainly not neurological, seeing as it responded in the expected way to this dangerous and ineffective psychotropic drug. Despite your no-doubt frequently-reliable instincts, mental health professionals nowadays do often cure memory deficits using drugs that are not normally prescribed for dementia. Many illnesses still are not usefully treatable, but that’s no reason to avoid treating those that are. It is our misfortune that we have no objective way, yet, to distinguish between those than to see which respond well to attempted treatments.
Seth: Because the models have been working well — i.e., predicting subsequent events accurately — for the past few years. In 2000, it was still possible to claim that the evidence didn’t demonstrate anything conclusively. Today there is enormously more evidence, and more kinds of evidence, and it overwhelmingly supports one side. You’re left grasping at straws, and it’s getting embarrassing.
Since the models neither incorporate nor predict cloud cover, the planet might leave the models behind if the atmosphere switches to a mode in which cloud cover is radically different. Chaotic systems are commonly near-stable only within a narrow range of conditions, and tend to lurch suddenly and dramatically to new modes in response to what seem like minor changes. The planet could equally as well plunge into a global ice age after a mode change triggered by initial warming. Such an outcome does not strike me as better. If we really trusted the models, we could drive the planet toward some preferred future condition. If you don’t trust the models, the only sane action is to avoid pushing the system out of its present, precarious balance.
In any case, the ongoing obliteration of reefs by carbonic acid poisoning is reason enough, all by itself, for heroic efforts to restrict carbon output. Consider the alternatives: (1) status quo, continued poisoning (mercury, sulfur, NOx, fouled fisheries, what-have-you), military extremism, and dependence on foreign political stability and largesse, or (2) clean wind and solar energy, with none of the above. What if global warming fizzles, and we got off our coal-and-oil habit “for nothing”? I’d rather live in world (2) either way. Avoiding global climatic catastrophe is a bonus.
“…mental health professionals nowadays do often cure memory deficits using drugs…” And not just nowadays — they’ve been doing it for years. Only the marketing is better today.
Nathan, I don’t see how we can get off our coal-and-oil habit “for nothing”. Every decision has a cost. There are no free lunches. In this case, imposing carbon emission limitations is likely to have a very heavy economic cost in developing countries, by severely limiting their economic growth. By limiting economic growth, we are prolonging the time they are shackled in poverty, which I would say is a very heavy cost. You might say that we (the developed world) should do all the R&D for renewable energy technology, then pass the benefits along to the developing world, but I doubt this can be done a) anytime soon, and b) without still imposing a high economic and therefore social cost to them. Of course the cost of doing nothing at all is that we do wreak havoc to the climate system, but there is still much uncertainty there, whereas the economic/social cost of reducing carbon is much more certain. On balance, it seems that the most prudent course of action is to wait a little longer for the science to solidify before acting one way or another.
Nathan, you believe the models “because the models have been working well — i.e., predicting subsequent events accurately — for the past few years.” I am unaware of this correct prediction. Could you tell me where I can learn about it?
I wouldn’t have put it the way that Nathan-no-relation did, hence my emphasis on postdiction, but quite a number of counterintuitive predictions have been borne out by the data. Consider https://bartonpaullevenson.com/ModelsReliable.html which claims, with references, that
Global Climate Models have successfully predicted:
* That the globe would warm, and about how fast, and about how
much.
* That the troposphere would warm and the stratosphere would cool.
* That nighttime temperatures would increase more than daytime
temperatures.
* That winter temperatures would increase more than summer temperatures.
* Polar amplification (greater temperature increase as you move
toward the poles).
* That the Arctic would warm faster than the Antarctic.
* The magnitude (0.3 K) and duration (two years) of the cooling
from the Mt. Pinatubo eruption.
* They made a retrodiction for Last Glacial Maximum sea surface
temperatures which was inconsistent with the paleo evidence, and better
paleo evidence showed the models were right.
* They predicted a trend significantly different and differently
signed from UAH satellite temperatures, and then a bug was found in
the satellite data. … …
And so on. Now, it’s q
Sean: By “for nothing”, I meant “to avert a catastrophe which later information indicated was not imminent”. Such an outcome might, for example, result from an (evidently delayed!) adjustment to changing conditions that increases the biosphere’s ability to sequester carbon. We might imagine marine plankton evolving (mystifyingly slowly) to grow thicker shells, which sink to the ocean bottom and remove carbon from circulation. I don’t know of any reason why that would occur. All the evidence we have indicates it’s not.
Remaining on coal and oil has imposed, and imposes, ruinous costs on the world not directly borne by extractors or purchasers. Only one of these is the ecological-network collapse and mass extinction that we see following global warming . Another is the ongoing obliteration of coral reefs by carbonic acid, and the fisheries that depend on them. Another is the pervasive distribution of mercury that had been safely sequestered underground. Another is the widespread tyranny and military oppression that ensures the elites’ exclusive access to carbon found under poor peoples’ land. (I could go on.) A carbon tax, applied scrupulously, would direct a tiny fraction of those costs back to users, and fund an efflorescence of new industries that have long been suppressed by the artificially low prices of oil and coal, soon eliminating such dependence and all of the costs.
We have already “waited a little longer”, again, and again, and again. All but 2 of some 2000 NAS members consider it solid already. Can you identify any condition that you would interpret to demonstrate the science has solidified? Are you holding out for unanimity?
Alex: Disingenuousness should be beneath us here.
I’d like to take the opportunity to explore why the double-blind trials always touted as the “gold standard” of drug effectiveness testing fail. It comes down to diagnostic categories. If symptom A has underlying causes M, N, O, P, …, of which N is seen, clinically, to be effectively treated with drug D, then the success of a trial depends entirely on how well you can distinguish N from M and O without resorting to D. If you cannot distinguish them with lab tests, then a trial must include a full population of sufferers of M, N, O, P, … It will only prove that D is ineffective for most of them, a fact you already knew all too well.
This difficulty with trials does not, in practice, cripple effective treatment, except insofar as it keeps useful compounds out of the pharmacopoeia. Psychiatrists and oncologists try D, E, F, … in succession, and when one is found to work, stop there. It’s a stone-age approach that reflects stone-age diagnostic capabilities, but the primitiveness of our present diagnostic capabilities should not keep patients from receiving treatments that are very often very effective.
Seth, all you need to do is step out of the echo chamber of the denialists, and you will find yourself deluged with compelling evidence. Bring a kayak.
sorry, not sure how that last was pre-submitted. Anyway I was going it say that:
Now, it’s quite true that this doesn’t give us a usefully testable prediction of the temperature of 2015, it gives us a range. So climate scientists do offer to place bets, but it’s not like betting that the Moon will be in its predicted position as of the year 2,000,000AD. But if you’re talking, as Nathan is and as I believe you are, about predictions used to validate a model — well, as I said, it’s been quite a while since I felt able to laugh at the climate models. (And as Nathan suggests, and as I think Tyler Cowen has said, “uncertainty is not your friend” in this issue; if I were confident that the models were correct, and that the real problem is the long-run problem, then I’d worry less about climate change than I do.)
(sorry for split comment.)
Nathan, you remind me of a surgeon I consulted who said there was plenty of evidence that the operation she recommended was a good idea. I could easily find the evidence using Google, she said. I could not find it. My mom, a medical librarian, could not find it. Confronted with these failures, the surgeon said she would find it. She never did. Let me make a prediction: You are never going to supply an example of the “successful prediction” you claim exists.
Tom, I looked at the page you linked to. The main prediction of the models is that there will be excessive warming, so I was especially curious about “global warming” predictions. The latest paper (of evidence) should be the best. Here is the latest paper that page gives:
Mann, M.E., Z. Zhang, M.K. Hughes, R.S. Bradley, S.K. Miller, S. Rutherford, and F. Ni 2008. “Proxy-based reconstructions of hemispheric and global surface temperature variations over the past two millennia.” Proceedings of the National Academies of Science 105, 13252-13257.
The title shows this isn’t about predictions, it is about fitting past data. Failing to understand the difference between prediction and fitting past data is so naive it seemed pointless to continue.
At any time — right now, 5 years ago, 10 years, 15 years ago — the people who work with those models and claim we should pay attention to their predictions could make/have made a set of predictions: next year, the year after that, and so on. Then, as time passed, we would have found out if the models predict correctly. The modelers haven’t done that. The more time that passes, the more glaring this failure becomes.
You are never going to supply an example of the “successful prediction” you claim exists
And my prediction: you will never look at any example without finding some way to ignore most of it, and re-interpret some detail in just such a way as to make you right all along. Will you wait until after the Amazonian basin is a desert, and insist that at no point was the evidence ever strong enough to act on? My question for you is the same as for Alex: what can you think of, that could happen within a year or three, that would oblige you to change your mind? Is any such thing even conceivable?
Ten years ago you could point to Mann’s statistical irregularities to proclaim his data too weak to demonstrate, alone, what is now well-supported. Today you’re reduced to promoting random offhand remarks as the “Best Argument Against Man-Made Global Warming”. Before, said warming was illusory. Now, it’s just not caused by CO2. Next will it be the fault of the ocean microfauna for failing to step up and absorb our excess? Where are we going from here?
Two points, I guess.
(1) Seth, you say The title shows this isn’t about predictions, it is about fitting past data. Failing to understand the difference between prediction and fitting past data is so naive it seemed pointless to continue. I do indeed wonder if it’s pointless to continue. If, say, Seth Roberts in his work on diet comes up with an appetite-model which implies a claim about the Paleolithic, and then somebody finds a cave in Iran with traces of ancient campfires which tests that claim, then I’m gonna list it as a “prediction”; the fires themselves were not in your future, but the connection of your model to that data was in your future, and that’s what counts epistemologically for evolutionary theory, for geology, for astrophysics or for climate change.
These millenia-old datasets, reported in the 2008 article you’re rejecting by title, were in the future for Arrhenius in 1896, for Broecker in 1975, and for the guys in between; those are the names in the “Models” column as having originated the general greenhouse-gas global warming prediction for which the 2008 article is among the “Evidence.” The compatibility of their models with that data strikes me as a genuine prediction-based test. (Of course the models are compatible with the rise in temperature with rising CO2 since the models were made, and that’s another.) And no, I don’t know that Levenson’s correct about this 2008 reference; I know about some of his references, but not this one. I’m just saying I don’t believe you can tell from the title that he’s using “prediction” improperly.
(2) Seth, you say that At any time — right now, 5 years ago, 10 years, 15 years ago — the people who work with those models and claim we should pay attention to their predictions could make/have made a set of predictions: next year, the year after that, and so on. Then, as time passed, we would have found out if the models predict correctly. The modelers haven’t done that.
So far as I can tell, that’s, well, that’s not quite right. The people who work with these models have said what the models predict, which is a little like predicting the water level as the tide comes in during a storm: things go up and down a lot, and there’s probably a storm surge adding a bias to your readings, so you cannot predict the exact level at any given time; prediction is rougher than that. It sounds like you’re demanding that they predict the waves; if that’s what you’re demanding, there’s not much point talking about it. I’m interested in knowing if the models (which make lots of predictions by which to be tested) are basically correct, because if so things will get gradually hotter and many people think that will be a disaster but I just think it will be a problem — if it gets hotter slowly, a solvable problem. So what predictions do they make? Well, the ones Levenson’s page lists… and others. Most famously I’m sure you’re aware of Hansen’s 1988 Scenarios A, B, and C with B being, he said, the most likely, and we now find, the closest to reality. I already linked to a discussion of that one, which has been dishonestly called a “failure” based on a false description (as described in the page I linked, and many other places) — but was it, honestly, a “success”? That depends on how you score it. From a prediction-evaluation point of view on the page I linked, I’d personally take most seriously Hansen’s model correctly projected amplified warming in the Arctic, as well as hot spots in northern and southern Africa, west Antarctica, more pronounced warming over the land masses of the northern hemisphere, etc. The spatial distribution of the warming is very close to his projections. That seems to me to speak to the underlying mechanism of the model. The exact amount is off, but not by much; the best description, as per that web page and other discussions I’ve seen, is simply that the climate sensitivity number Hansen used (that’s one parameter, one number) was within the currently-accepted range, but at the high end, while the number that would have made his curve come out “right” is slightly lower, near the middle of that currently-accepted range.
So, looking at that, I look at what the critics (besides the dishonest ones) say about it, and I look for somebody who finds a way to justify an extremely low climate sensitivity number with some other model — something that will also produce amplified warming in the Arctic, etc. And so far, I don’t find it. So far, I’m left with the simplest description being “Hansen’s climate sensitivity number was a bit high, but it looks like he got things pretty much right.” That, to me, is a fairly reassuring description, because it suggests that we may not have real climate disasters before we can develop the technology to warm or cool the Earth as much as we want. I’m a Moore’s-Law sort of geek. But it’s not completely reassuring, because all models are wrong, and I do not feel that uncertainty in this matter is my friend.
Tom, as the term Little Ice Age implies, global temperatures have been rising for a while. When Hansen predicts they will continue to rise at roughly the same rate — and they do — I hope you can see that that is no success for the model. You have to improve on common sense to be taken seriously. My weight-control theory led me to find that drinking sugar water can cause weight loss. That’s wasn’t common sense — it was the opposite of what everyone believed. That’s an example of what should increase belief in a theory.
I didn’t expect my prediction to be borne out so soon and so thoroughly,
Nathan Myers wrote: “My question for… Alex [is]: what can you think of, that could happen within a year or three, that would oblige you to change your mind? Is any such thing even conceivable?”
I already changed my mind, Nathan. When I worked on the antidepressant team at a pharmaceutical company (in the 1990s), I had a largely positive view of antidepressants and other psychotropic drugs. But I also learned some troubling things, and I maintained my interest in the subject long after I moved into a different line of work.
After reading books like these, and articles like this one, I changed my mind.
Incidentally, Robert Whitaker, who is currently the leading skeptic of psychotropic drugs, was recently invited to present at Grand Rounds at Massachusetts General Hospital (Harvard’s teaching hospital). He probably didn’t change many minds, but his very presence there is a hopeful sign to me that maybe the tide is turning.
Well, Seth, you have done very valuable work in collecting surprises and turning them into theories. I was intrigued when I read Gelman’s blog about your work, tried the diet myself a few months later, wrote about it more than a year after that at https://mistakesbytjm.blogspot.com/2007/02/exercise-diet-self-experimentation.html and I’m sure that you have an important piece of the puzzle. What impressed me in the end about your model, the “flavor-calorie theory”, was the incidental predictions: you could lose weight with sugar water, but your model of that phenomenon then predicted that you could lose weight with unflavored oil, nose-clipping, crazy-spicing, etc, and they all worked — at least for a lot of people, to some extent, with unpredictable individual variations as noted at the forum and random places on the internet. It’s clear that the overall picture is complicated (and includes some Taubes-style stuff), but this belongs in it. If I Were In Charge, the main federal involvement in medicine would not be regulation but data-collection, and Roberts’ flavor-calorie theory would be one of the initial targets. As I said at https://mistakesbytjm.blogspot.com/2007/10/self-experimentation-dental-care-health.html “I’m very glad there are people like Seth Roberts in the universe.” Besides those items, I started this morning with 45 minutes, just short of five miles, on an elliptical machine with flash cards. Modern Greek flash cards, 45 elliptical minutes every morning — I can’t do it on a treadmill because my ankles start hurting after a few days. When I go back to Greece this summer (for a 2nd granddaughter’s christening, my daughter-in-law being now Greek-American) I will do better than I did before, and you’ve made it much easier. So — thanks.
But as I said, what made me take the flavor-calorie model seriously was the collection of incidental predictions; it wasn’t just saying that you’d made an observation about sugar water and more sugar water was better. Most useful models touch the world in many places. Your comment that we’re coming out of an ice age, warming is common sense, is what I heard in the 80s from at least one of the scientists in my family (we run to PhDs, dunno why). He changed his mind when the trend (see https://en.wikipedia.org/wiki/Global_warming ) went further faster than he was expecting, more than ten years ago. I’ve heard “global warming has stopped, we’re going to be cooling now” claims since about that same time. Hansen published his curves in 1988, and that was controversial; do you have a comparable prediction from the same time-period, from “common sense” or geophysical modeling? I didn’t have a strong sense of how much of a trend to expect, overall, but at about the same time I changed my own mind because there were a lot of incidental predictions, such as those I commented on and then quoted above (Arctic, etc). We’re not talking about a single number-sequence; we’re talking about a model, one which touches the world in many places.
It sounds as if you are saying that you don’t intend to check climate models by their incidental predictions because those aren’t central, and you don’t intend to check climate models by the central warming prediction because you were sort of expecting some warming anyway. And that sounds as if you simply don’t want to think about it; you’ll wait until…well, you’ll wait, never mind the “until”. After all, it was below zero here in upstate NY this morning; surely the world’s not warming. And that’s okay; I’m still glad there are people like Seth Roberts in the universe.
Tom, I’m pleased you’ve been able to use my treadmill/learning discovery. It’s made a big difference to me, too.
About testing climate models: I wrote a paper with Hal Pashler about how to test complex models. That paper was about complex models in psychology but of course the same ideas apply in other areas. I’m saying here what I said there.
In psychology, many professors (modelers) at respected universities were essentially cheating by exploiting a gap in understanding. They passed off a test that was easy to pass as a test that was hard to pass and thereby got their models published. The gap in understanding was that the reviewers of their papers didn’t understand how to test a theory; they didn’t understand what is and isn’t a useful test. Proponents of the climate models we are talking about constantly do the same thing: treat an easy test as if it were a hard test. Their models have passed easy tests but not hard tests, as far as I can tell. It’s as if they hand me Monopoly money and expect me to treat it like real money. When I say “this money hasn’t been tested” they say: Look, it’s the right size!
I guess we’re back where we started, which is not a bad place to be: you started with a quote including “it is only when they add CO2 influences to the models that they can explain recent global warming” and now you say the tests they pass are easy. Presumably you mean that it is easy to come up with a model that passes the same tests but doesn’t depend on the CO2 sensitivity. A model, as Levenson was noting, that generates the temperature distributions across different regions and different layers of the atmospheres and different day/night and winter/summer times, a model that may not be fully precise but was precise enough to show up the errors in Roy Spencer’s satellite data and the paleo evidence Levenson notes and … ? I wouldn’t know how to do it, but I’m just a programmer and ex-computer science prof. The tests they’ve passed have looked pretty hard to me. Of course there are many tests they haven’t passed yet, and may never pass (weather is fundamentally unpredictable unless it is someday actually controlled, in the sense that initial-measurement errors necessarily exist and necessarily grow exponentially; I have not seen anything to give me a strong opinion one way or the other on whether or not climate, i.e. average weather, has strange-attractor problems of the same sort. My feeble opinion is no.) If somebody has come up with such a model, let them put it online as open source, like the GISS climate model, and let Hansen, Schmidt et al have a whack at it: I’ll buy popcorn. Meanwhile, I am facing the fact that after more than 20 years I haven’t seen such a model, and I’ve been waiting for one. I predict that one won’t be forthcoming. Will I bet my life on that? Am I absotively posilutely sure there isn’t one? Of course not. It may exist. I think the climate models are pretty good; I’m pretty sure they’re the best guess we’ve got; but it’s not a perfect world, and being models they are certainly somewhat wrong, and they might be seriously wrong.
That’s not reassuring.
As Tyler Cowen said in a post called “My views on global warming”:I do not much trust climate models. Perhaps I have spent too much time doing macro, and the experience carries over. Nonetheless uncertainty about final effects gives us more to worry about, not less. It is the worst-case scenarios for global warming which worry me, not the middling scenarios. Variance is our enemy in this matter. So it goes.
(It occurs to me that I really ought to have been writing code all this time, even if I got through the last deadline fine and the next one is a ways off. Anyway, thanks for lots of things.)
I would like to know Seth’s thoughts on the subject of confirmation bias.
Alex: You seem to be saying that as a welcome consequence of this turned tide, my psychiatrist would not be allowed to prescribe the medicine that restored my ability to do work and remember phone numbers.
Nathan: I’m in favor of legalizing all drugs (whether I happen to like them or not). So under my system, you’d be able to prescribe the drug for yourself.
Alex: I would never, ever have thought of using that particular compound for that purpose. Chalk one up for the pros.
“You are never going to supply an example of the ‘successful prediction’ you claim exists.”
I agree. I have had the “successful prediction” argument many times with warmists. Very frequently, their example of a successful prediction is actually a back-casting. A post-diction, if you will.
And usually these back-casts are a beautiful fit to past data. Which actually makes them even more suspicious.
apparently u haven’t heard many arguments against man made global warming…you bad.
Are we kidding here? Ok Co2 levels are at levels never recorded before. Co2 IS a greenhouse gas, but not the biggest one. Thats Methane. Guess what Methane is also at levels NEVER SEEN BEFORE! Your right, we cant accurately predict whats going to happen exactly. That should scare the crap out of you. The planet will respond. It always has, it always will. The responce WILL move the climate out of our comfort zone. There is no argument against Man Made Climate Change. There is scientific conceses, it is happening. We did it. The question is not will it happen, but just how bad it will get. If we stop now, perhaps not as bad as we can fear, if we keep giving out misinformation like this blog then it will get very bad indeed. The Cretacous was not a time human would like to live in. That was a natral climate. We have pushed the balance past that and into UNKOWN teritory. If that fact alond does not scare the pants off you then you are not paying attention.